Microsoft limits Bing chats to 5 questions per session, experts say angry chatbot 'just mimicking human conversations'
Crazy tales of exchanges with the Microsoft Bing chatbot have captured attention this week including the AI issuing threats and telling of desires to steal nuclear code, create a deadly virus, and even profess love.
- Feb 18, 2023,
- Updated Feb 18, 2023 9:43 AM IST
Microsoft said on Friday that it will limit some of the chat functionality on its new Bing search engine powered by generative artificial intelligence (AI). The questions will now be limited to five per session and 50 questions per day.
"As we mentioned recently, very long chat sessions can confuse the underlying chat model in the new Bing. To address these issues, we have implemented some changes to help focus the chat sessions," Microsoft said in the blog post.
Meanwhile, weird and crazy tales of exchanges with the Microsoft Bing chatbot have captured attention this week including the AI issuing threats and telling of desires to steal nuclear code, create a deadly virus, or to be alive and even profess love.
The new Bing, which has a waitlist of millions of people for access, is a potentially lucrative opportunity for Microsoft. The company said during an investor and press presentation last week that every percentage point of market share it gains in the search advertising market could bring in another $2 billion of ad revenue. The company also announced a new multiyear, multibillion-dollar investment with OpenAI in January this year.
Experts on AI chatbots
Experts are now coming out and trying to explain the behaviour of AI chatbots.
Graham Neubig, an associate professor at Carnegie Mellon University's language technologies institute said, "I think this is basically mimicking conversations that it's seen online."
"So once the conversation takes a turn, it's probably going to stick in that kind of angry state, or say 'I love you' and other things like this, because all of this is stuff that's been online before," Neubig added.
Programmer Simon Willison talked about how these AI models don't have a concept of truth in a blog post. "Large language models have no concept of 'truth' -- they just know how to best complete a sentence in a way that's statistically probable based on their inputs and training set," he said.
The rectification and normalisation of such AI chatbots will take time according to experts as it requires a lot of human feedback.
"Addressing this requires a lot of effort and a lot of human feedback, which is also the reason why we chose to restrict ourselves for now to business uses and not more conversational ones," Laurent Daudet, co-founder of French AI company LightOn, told AFP.
Also Read
Google showcases what its ChatGPT competitor ‘Bard’ can do
New YouTube CEO Neal Mohan could’ve been with Twitter, but Google paid him $100 million to stay
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine
Microsoft said on Friday that it will limit some of the chat functionality on its new Bing search engine powered by generative artificial intelligence (AI). The questions will now be limited to five per session and 50 questions per day.
"As we mentioned recently, very long chat sessions can confuse the underlying chat model in the new Bing. To address these issues, we have implemented some changes to help focus the chat sessions," Microsoft said in the blog post.
Meanwhile, weird and crazy tales of exchanges with the Microsoft Bing chatbot have captured attention this week including the AI issuing threats and telling of desires to steal nuclear code, create a deadly virus, or to be alive and even profess love.
The new Bing, which has a waitlist of millions of people for access, is a potentially lucrative opportunity for Microsoft. The company said during an investor and press presentation last week that every percentage point of market share it gains in the search advertising market could bring in another $2 billion of ad revenue. The company also announced a new multiyear, multibillion-dollar investment with OpenAI in January this year.
Experts on AI chatbots
Experts are now coming out and trying to explain the behaviour of AI chatbots.
Graham Neubig, an associate professor at Carnegie Mellon University's language technologies institute said, "I think this is basically mimicking conversations that it's seen online."
"So once the conversation takes a turn, it's probably going to stick in that kind of angry state, or say 'I love you' and other things like this, because all of this is stuff that's been online before," Neubig added.
Programmer Simon Willison talked about how these AI models don't have a concept of truth in a blog post. "Large language models have no concept of 'truth' -- they just know how to best complete a sentence in a way that's statistically probable based on their inputs and training set," he said.
The rectification and normalisation of such AI chatbots will take time according to experts as it requires a lot of human feedback.
"Addressing this requires a lot of effort and a lot of human feedback, which is also the reason why we chose to restrict ourselves for now to business uses and not more conversational ones," Laurent Daudet, co-founder of French AI company LightOn, told AFP.
Also Read
Google showcases what its ChatGPT competitor ‘Bard’ can do
New YouTube CEO Neal Mohan could’ve been with Twitter, but Google paid him $100 million to stay
For Unparalleled coverage of India's Businesses and Economy – Subscribe to Business Today Magazine